import numpy as np
import pandas as pd
import geopandas as gpd
import altair as alt
import requests
from tqdm import tqdm
import datetime

0. Data Preparation

0.1. Vaccination in ukraine data aggregation.

vaccination_data_url = 'https://health-security.rnbo.gov.ua/api/vaccination/process/table?perPage=125&page={}&orderBy=date&dose=all&distributionBy=vaccine&vaccines=&period=daily&aggregateBy=regions'

def get_vaccination_data(page_num: int) -> pd.DataFrame:
    url = vaccination_data_url.format(page_num)
    df = pd.DataFrame(requests.get(url).json()['data'])
    df = df.drop(columns=['SarsCov2_nRVv3', 'SarsCov2_nRVv3_pct'])
    df = df.rename(columns={'name': 'region'})
    df = df.fillna(0)
    df.date = pd.to_datetime(df.date, infer_datetime_format=True)  
    return df

vaccination_ukraine_df = get_vaccination_data(1)
for i in tqdm(range(2, 37)):
    vaccination_ukraine_df = vaccination_ukraine_df.append(get_vaccination_data(i))

100%|███████████████████████████████████████████| 35/35 [02:24<00:00,  4.14s/it]

vaccination_ukraine_df.head()

	date	region	Moderna	AstraZeneca	Moderna_pct	AstraZeneca_pct	Pfizer-BioNTech	Pfizer-BioNTech_pct	Sinovac (CoronaVac)	Sinovac (CoronaVac)_pct	Johnson & Johnson	Johnson & Johnson_pct
0	2021-12-22	Київська область	154	215	2.81	3.92	2780	50.73	2331	42.54	0.0	0.0
1	2021-12-22	Луганська область	409	58	18.78	2.66	972	44.63	739	33.93	0.0	0.0
2	2021-12-22	Волинська область	359	115	10.49	3.36	2248	65.71	699	20.43	0.0	0.0
3	2021-12-22	Миколаївська область	251	25	10.07	1.00	1251	50.20	965	38.72	0.0	0.0
4	2021-12-22	Черкаська область	338	122	10.77	3.89	1350	43.03	1327	42.30	0.0	0.0

vaccination_ukraine_df.dtypes

date                       datetime64[ns]
region                             object
Moderna                             int64
AstraZeneca                         int64
Moderna_pct                       float64
AstraZeneca_pct                   float64
Pfizer-BioNTech                     int64
Pfizer-BioNTech_pct               float64
Sinovac (CoronaVac)                 int64
Sinovac (CoronaVac)_pct           float64
Johnson & Johnson                 float64
Johnson & Johnson_pct             float64
dtype: object

vaccination_ukraine_df.describe()

	Moderna	AstraZeneca	Moderna_pct	AstraZeneca_pct	Pfizer-BioNTech	Pfizer-BioNTech_pct	Sinovac (CoronaVac)	Sinovac (CoronaVac)_pct	Johnson & Johnson	Johnson & Johnson_pct
count	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000
mean	591.856889	624.592889	11.108153	11.851831	2527.220222	44.897076	1873.640222	31.929253	4.508889	0.169304
std	1307.180669	743.954089	16.031511	9.428439	2523.799513	15.915060	2028.583628	14.992706	83.140434	3.132487
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000
25%	0.000000	150.000000	0.000000	4.887500	1020.000000	35.757500	547.750000	21.725000	0.000000	0.000000
50%	26.500000	400.000000	0.675000	10.195000	1793.500000	45.510000	1241.000000	30.890000	0.000000	0.000000
75%	705.000000	801.250000	19.335000	16.735000	3194.250000	55.042500	2495.500000	39.890000	0.000000	0.000000
max	15998.000000	6936.000000	98.300000	100.000000	21139.000000	100.000000	19105.000000	100.000000	2300.000000	80.910000

0.2. Covid statistics in ukraine data aggregation.

covid_stats_data_url = 'https://api-covid19.rnbo.gov.ua/data?to={}'
def get_covid_stats_data(date: datetime.date) -> pd.DataFrame:
    url = covid_stats_data_url.format(date.isoformat())
    df = pd.json_normalize(requests.get(url).json()['ukraine'])
    df['date'] = date.isoformat()
    df = df.drop(columns=['id', 'country'])
    df = df.rename(columns={'label.uk': 'region', 'label.en': 'region_en'})
    df = df.fillna(0)
    df.date = pd.to_datetime(df.date, infer_datetime_format=True)  
    return df


today_date = datetime.date.today()
covid_stats_ukraine_df = get_covid_stats_data(today_date)
for i in tqdm(range(1, 180)):
    covid_stats_ukraine_df = covid_stats_ukraine_df.append(get_covid_stats_data(today_date - datetime.timedelta(days=i)))

100%|█████████████████████████████████████████| 179/179 [01:29<00:00,  2.00it/s]

covid_stats_ukraine_df.head()

	confirmed	deaths	recovered	existing	suspicion	lat	lng	delta_confirmed	delta_deaths	delta_recovered	delta_existing	delta_suspicion	region_en	region	date
0	111059	3029	99285	8745	59638	48.920517	28.685484	201	13	1046	-858	0	Vinnytsia	Вінницька область	2021-12-23
1	102685	2148	97524	3013	49951	51.190885	24.870603	221	4	289	-72	0	Volynska	Волинська область	2021-12-23
2	240883	8707	227545	4631	96446	48.301142	34.844169	231	21	163	47	0	Dnipropetrovska	Дніпропетровська область	2021-12-23
3	165359	5041	154439	5879	60058	48.047635	37.674086	531	13	789	-271	0	Donetska	Донецька область	2021-12-23
4	139486	3204	133023	3259	72918	50.639546	28.475602	243	8	542	-307	0	Zhytomyrskа	Житомирська область	2021-12-23

covid_stats_ukraine_df.dtypes

confirmed                   int64
deaths                      int64
recovered                   int64
existing                    int64
suspicion                   int64
lat                       float64
lng                       float64
delta_confirmed             int64
delta_deaths                int64
delta_recovered             int64
delta_existing              int64
delta_suspicion             int64
region_en                  object
region                     object
date               datetime64[ns]
dtype: object

covid_stats_ukraine_df.describe()

	confirmed	deaths	recovered	existing	suspicion	lat	lng	delta_confirmed	delta_deaths	delta_recovered	delta_existing	delta_suspicion
count	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.000000	4500.0
mean	106490.874222	2542.880889	97094.810222	6853.183111	73654.800000	49.172537	30.511702	310.410000	9.314444	273.655111	27.440444	0.0
std	53512.977875	1431.192910	47307.370362	8930.202204	33647.792136	1.360405	4.423417	394.763285	13.636438	455.353751	411.358887	0.0
min	20610.000000	837.000000	19096.000000	46.000000	17789.000000	46.518829	23.275854	0.000000	0.000000	0.000000	-4949.000000	0.0
25%	72995.250000	1511.750000	69254.750000	529.000000	57582.000000	48.301142	26.391115	28.000000	0.000000	16.000000	-14.000000	0.0
50%	91066.000000	2076.500000	86546.500000	2880.500000	72918.000000	49.260159	30.602712	148.000000	3.000000	70.000000	16.000000	0.0
75%	135761.250000	3044.250000	125438.750000	10569.500000	86388.000000	50.299214	33.778584	452.250000	13.000000	333.250000	138.250000	0.0
max	325257.000000	8707.000000	305034.000000	51211.000000	182459.000000	51.355018	39.018116	2616.000000	144.000000	5309.000000	2028.000000	0.0

covid_and_vaccination_ukraine_df = vaccination_ukraine_df.merge(covid_stats_ukraine_df, left_on=['region', 'date'], right_on=['region', 'date'])
covid_and_vaccination_ukraine_df

	date	region	Moderna	AstraZeneca	Moderna_pct	AstraZeneca_pct	Pfizer-BioNTech	Pfizer-BioNTech_pct	Sinovac (CoronaVac)	Sinovac (CoronaVac)_pct	...	existing	suspicion	lat	lng	delta_confirmed	delta_deaths	delta_recovered	delta_existing	delta_suspicion	region_en
0	2021-12-22	Київська область	154	215	2.81	3.92	2780	50.73	2331	42.54	...	9930	101468	50.299214	30.455333	354	22	1309	-977	0	Kyivska
1	2021-12-22	Луганська область	409	58	18.78	2.66	972	44.63	739	33.93	...	1587	18754	48.985624	39.018116	77	6	279	-208	0	Luhanska
2	2021-12-22	Волинська область	359	115	10.49	3.36	2248	65.71	699	20.43	...	3085	49951	51.190885	24.870603	376	7	712	-343	0	Volynska
3	2021-12-22	Миколаївська область	251	25	10.07	1.00	1251	50.20	965	38.72	...	3666	54827	47.438413	31.778998	219	8	268	-57	0	Mykolaivska
4	2021-12-22	Черкаська область	338	122	10.77	3.89	1350	43.03	1327	42.30	...	7891	61376	49.260159	31.352536	309	16	925	-632	0	Cherkaska
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
3570	2021-06-27	Херсонська область	0	30	0.00	8.90	0	0.00	307	91.10	...	445	25112	46.673984	33.535980	15	0	3	12	0	Khersonska
3571	2021-06-27	Закарпатська область	0	50	0.00	7.54	0	0.00	613	92.46	...	286	57582	48.403718	23.275854	3	0	21	-18	0	Zakarpatska
3572	2021-06-27	Рівненська область	0	10	0.00	2.36	348	82.08	66	15.57	...	294	75027	51.039594	26.391115	9	0	11	-2	0	Rivnenska
3573	2021-06-27	Миколаївська область	0	30	0.00	8.85	108	31.86	201	59.29	...	574	54827	47.438413	31.778998	22	2	13	7	0	Mykolaivska
3574	2021-06-27	Сумська область	0	0	0.00	0.00	30	4.53	632	95.47	...	554	65456	51.106120	34.123810	15	0	25	-10	0	Sumska

3575 rows × 25 columns

0.3. Ukraine map preparation.

ukraine = gpd.read_file('data/ukraine.json')
ukraine = ukraine[['NAME_1', 'geometry']]
ukraine = ukraine.rename(columns={'NAME_1': 'region'})
ukraine.head()

	region	geometry
0	Cherkasy	MULTIPOLYGON (((31.32614 48.74507, 31.31716 48...
1	Chernihiv	MULTIPOLYGON (((33.09283 50.50966, 33.09261 50...
2	Chernivtsi	MULTIPOLYGON (((24.93280 47.72794, 24.93301 47...
3	Crimea	MULTIPOLYGON (((33.79291 44.39153, 33.79465 44...
4	Dnipropetrovs'k	MULTIPOLYGON (((33.93176 47.48407, 33.92332 47...

alt.Chart(ukraine).project().mark_geoshape(
    fill = 'lightgray', stroke = 'white', strokeWidth = 1
).properties(
    width = 800, height = 600
).encode().configure_view(
    strokeWidth = 0
)

covid_stats_ukraine_df.region_en.unique()

array(['Vinnytsia', 'Volynska', 'Dnipropetrovska', 'Donetska',
       'Zhytomyrskа', 'Zakarpatska', 'Zaporizka', 'Ivano-Frankivska',
       'Kyivska', 'Kirovohradska', 'Luhanska', 'Lvivska', 'Mykolaivska',
       'Kyiv', 'Odeska', 'Poltavska', 'Rivnenska', 'Sumska',
       'Ternopilska', 'Kharkivska', 'Khersonska', 'Khmelnytska',
       'Cherkaska', 'Chernivetska', 'Chernihivska'], dtype=object)

ukraine_geofaset_grid = pd.DataFrame(data={
  'name': [
      "Volynska", "Kyiv", "Chernivetska", "Sumska", "Kyivska", "Zhytomyrskа", "Luhanska", "Lvivska", "Ternopilska", "Rivnenska",
      "Poltavska", "Kharkivska", "Zakarpatska", "Donetska", "Ivano-Frankivska", "Khmelnytska", "Vinnytsia", "Cherkaska", "Kirovohradska",
      "Dnipropetrovska", "Mykolaivska", "Khersonska", "Chernivetska", "Odeska", "Zaporizka", "Crimea"],
  'row': [1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 5],
  'col': [2, 5, 6, 7, 5, 4, 8, 1, 2, 3, 6, 7, 1, 8, 2, 3, 4, 5, 6, 7, 5, 6, 2, 4, 7, 5],
})
 
ukraine_geofaset_grid

	name	row	col
0	Volynska	1	2
1	Kyiv	1	5
2	Chernivetska	1	6
3	Sumska	1	7
4	Kyivska	2	5
5	Zhytomyrskа	2	4
6	Luhanska	2	8
7	Lvivska	2	1
8	Ternopilska	2	2
9	Rivnenska	2	3
10	Poltavska	2	6
11	Kharkivska	2	7
12	Zakarpatska	3	1
13	Donetska	3	8
14	Ivano-Frankivska	3	2
15	Khmelnytska	3	3
16	Vinnytsia	3	4
17	Cherkaska	3	5
18	Kirovohradska	3	6
19	Dnipropetrovska	3	7
20	Mykolaivska	4	5
21	Khersonska	4	6
22	Chernivetska	4	2
23	Odeska	4	4
24	Zaporizka	4	7
25	Crimea	5	5

ukraine_geofaset_base = alt.Chart(ukraine_geofaset_grid).project().mark_square(
    fill = 'lightgray', size=8000
).properties(
    width = 800, height = 600
).encode(
    x = alt.X('col:O'),
    y = alt.Y('row:O'),
).configure_axis(
    grid=False, domain=False, labels=False, ticks=False, title=None
).configure_view(
    strokeWidth = 0
)


ukraine_geofaset_base

Now, we are ready to start analyzing existing data and create visualization. However, for that reason we firstly need to set the ground questions that we want to analyze in this work.

1. See the general statictics of covid spread in Ukraine (for the last 6 month).
2. Discover the vaccination tendends in Ukraine (for the last 6 month).

1. Covid statistics analysis.

1.1 General statictics.

newest_date_covid_stats_ukraine_df = covid_stats_ukraine_df[covid_stats_ukraine_df.date == covid_stats_ukraine_df.date.max()]
newest_date_covid_stats_ukraine_df = newest_date_covid_stats_ukraine_df[['recovered', 'existing', 'suspicion', 'deaths']].sum()
newest_date_covid_stats_ukraine_df = pd.DataFrame({'status': newest_date_covid_stats_ukraine_df.index, 'value': newest_date_covid_stats_ukraine_df.values})

alt.Chart(newest_date_covid_stats_ukraine_df).mark_bar(size=60).encode(
    x=alt.X('value:Q', title='Number of people'),
    color=alt.Color('status:N', title='Status'),
    tooltip=[
        alt.Tooltip('status:N', title='Status'),
        alt.Tooltip('value:Q', title='Number of people'),
   ]
).properties(
    width=800,
    height=70,
)

This visualization shows the newest daily updated destribution of covid case.

Alterntive approach for visualizing this data would be to use the standart barplot or pieplot (for some reasons isn't working), either of these solutions will work good in this situation.

alt.Chart(covid_stats_ukraine_df).transform_fold(
    ['recovered', 'existing', 'suspicion', 'deaths'],
).mark_area().encode(
    x=alt.X('month(date):O', title='Month'),
    y=alt.Y('value:Q', aggregate='sum', title='Covid patients status propostion'),
    color=alt.Color('key:N', title='Patient Status'),
    order=alt.Order('value:Q', aggregate='sum', sort='ascending'),
    tooltip=[
        alt.Tooltip('key:N', title='Status'),
        alt.Tooltip('value:Q', aggregate='sum', title='Number of people'),
        alt.Tooltip('month(date):O', title='Month'),
   ]
).properties(width=800, height=500)

This viualization was made to compare the dynamic of covid cases registration through last 6 month, and show a new covid wave that happened this autumn.

Alternative approache was to use bar plot with normalization to simplify the plot a bit, by this areaplot hytlight better the new covid way effect.

1.2. Confirmed cases statistics comparisons by regions.

confirmed_case_by_month_and_region = covid_and_vaccination_ukraine_df.groupby([pd.Grouper(key='date', freq='1M'), 'region'])[['confirmed']].sum().reset_index() 
confirmed_case_by_month_and_region = confirmed_case_by_month_and_region.merge(covid_and_vaccination_ukraine_df, left_on=['region', 'date'], right_on=['region', 'date'])
confirmed_case_by_month_and_region = confirmed_case_by_month_and_region[['date', 'region', 'lat', 'lng', 'confirmed_x']]
confirmed_case_by_month_and_region = confirmed_case_by_month_and_region.rename(columns={'confirmed_x': 'confirmed'})
confirmed_case_by_month_and_region

	date	region	lat	lng	confirmed
0	2021-06-30	Івано-Франківська область	48.701932	24.619344	346619
1	2021-06-30	Волинська область	51.190885	24.870603	248528
2	2021-06-30	Вінницька область	48.920517	28.685484	284306
3	2021-06-30	Дніпропетровська область	48.301142	34.844169	542223
4	2021-06-30	Донецька область	48.047635	37.674086	362328
...	...	...	...	...	...
145	2021-11-30	Хмельницька область	49.507456	26.929458	2654143
146	2021-11-30	Черкаська область	49.260159	31.352536	2267734
147	2021-11-30	Чернівецька область	48.268275	25.976996	2280147
148	2021-11-30	Чернігівська область	51.355018	32.004509	1707910
149	2021-11-30	м. Київ	50.448668	30.547911	5741305

150 rows × 5 columns

background = alt.Chart(ukraine).project().mark_geoshape(
    fill = 'skyblue', stroke = 'white', strokeWidth = 1
).properties(
    width = 800, height = 600
).encode()

def timestamp(t):
    return pd.to_datetime(t).timestamp() * 1000

slider = alt.binding_range(
    step=30 * 24 * 60 * 60 * 1000, # 30 days in milliseconds
    min=timestamp(confirmed_case_by_month_and_region.date.min()),
    max=timestamp(confirmed_case_by_month_and_region.date.max()),
    name='Select Month:'
)

select_date = alt.selection_single(
    fields=['date'],
    bind=slider,
    init={'date': timestamp(confirmed_case_by_month_and_region.date.min())},
    name='slider')

points_base = alt.Chart(confirmed_case_by_month_and_region).encode(
    longitude=alt.Longitude('lng:Q'),
    latitude=alt.Latitude('lat:Q')
)

points = points_base.mark_circle().encode(
    size=alt.Size(
        'confirmed:Q',
        title='Number of confirmed cases',
        scale=alt.Scale(domain=[
            confirmed_case_by_month_and_region.confirmed.min(), confirmed_case_by_month_and_region.confirmed.max()
        ], range=[300, 2500])
    ),
    color=alt.Color('confirmed:Q', scale=alt.Scale(scheme='orangered', domain=[
            confirmed_case_by_month_and_region.confirmed.min(), confirmed_case_by_month_and_region.confirmed.max()
        ])),
    tooltip=[
        alt.Tooltip('region:N', title='Region'),
        alt.Tooltip('confirmed:Q', title='Number of confirmed cases')
    ],
    
).transform_filter(
    "(year(datum.date) == year(slider.date[0])) && "
    "(month(datum.date) == month(slider.date[0]))"
).add_selection(select_date)

text_labels = points_base.mark_text().encode(
    text=alt.Text('region:N'),
)

(background + points).configure_view(
    strokeWidth = 0
)

This viualization is designed to show the increase of new confirmed covid case and how quickly they increase throught the last 6 month.

However, the main disadvantage of this approach is that the confirmed case are not normalized by the region population and becaue of that bigger and pore populated regions gets more attantion on the map.

The second problem with this visualization is pour UI of standard HTML range that are used by Altair - the right preview of range value is bad (since the only way that worked for me is using timestamps) and the range lack of the stage for different month which creates a bad UX. Alternative approach here would be to use the radio select buttons, but that is not quite the expereince I would like to achive.

1.3. Covid cases status proportions comparisons by regions

region_base = alt.Chart(covid_stats_ukraine_df).mark_bar().transform_fold(
    ['recovered', 'existing', 'suspicion', 'deaths'],
).encode(
    x=alt.X('month(date):O', title='Month'),
    y=alt.Y('value:Q', aggregate = 'max', stack='normalize', title=None),
    color=alt.Color('key:N', title='Status'),
    order=alt.Order('value:Q', aggregate='sum', sort='ascending'),
    tooltip=[
        alt.Tooltip('key:N', title='Status'),
        alt.Tooltip('value:Q', aggregate='max', title='Number of people'),
        alt.Tooltip('month(date):O', title='Month'),
   ],
   detail = alt.Detail('region_en:N')
).properties(width=120, height=120)

region_base_title = alt.Chart(covid_stats_ukraine_df).mark_text(dy = -70).encode(
    text = alt.Text('region_en:N')
)

region_base += region_base_title

region_base.transform_filter(alt.datum.region_en == 'Vinnytsia')

empty_cell = alt.Chart(
    alt.Data(
        values=[{'x': covid_stats_ukraine_df.date.min(), 'y': 1}, {'x': covid_stats_ukraine_df.date.max(), 'y': 0}]
    )
).mark_point(
    opacity=0
).encode(
    x = alt.X('x:T', axis=alt.Axis(title=None, labelOpacity=0, tickOpacity=0, domainOpacity=0, grid=False)),
    y = alt.Y('y:Q', axis=alt.Axis(title=None, labelOpacity=0, tickOpacity=0, domainOpacity=0, grid=False))
).properties(width=120, height=120)

empty_cell

chart = alt.vconcat()

for row in ukraine_geofaset_grid.row.unique():
    chart_row = alt.hconcat()
    for col in range(1, 9):
        if col in ukraine_geofaset_grid[ukraine_geofaset_grid.row == row].col.unique():
            region_en = ukraine_geofaset_grid[(ukraine_geofaset_grid.row == row) & (ukraine_geofaset_grid.col == col)].name.values[0]
            
            if region_en in covid_stats_ukraine_df.region_en.unique():
                chart_row |= region_base.transform_filter(alt.datum.region_en == region_en)
            else:
                chart_row |= empty_cell
        else:
            chart_row |= empty_cell
    chart &= chart_row

chart.properties(
    title='Covid cases status proportions'
).configure_title(
    fontSize=30,
    anchor='middle',
    dy = -20
).configure_view(
    strokeWidth = 0
)

This visualization is a bit similar to the one used in 1.1 section, but with the difference that now all plots are normalized. That's done for vanishing the difference in regions population, but showing that for all the region the increase in the new covid case can be seen in the autumn months.

Alternatively, if we decide to how only the trends in the existing confirmed covid case, then we can use standart line plot and show only existing covid case, but again due to the lack of cities populations the visialization results would be unfair to the smaller cities.

2. Vaccination statistics analysis.

vaccination_ukraine_df.head()

	date	region	Moderna	AstraZeneca	Moderna_pct	AstraZeneca_pct	Pfizer-BioNTech	Pfizer-BioNTech_pct	Sinovac (CoronaVac)	Sinovac (CoronaVac)_pct	Johnson & Johnson	Johnson & Johnson_pct
0	2021-12-22	Київська область	154	215	2.81	3.92	2780	50.73	2331	42.54	0.0	0.0
1	2021-12-22	Луганська область	409	58	18.78	2.66	972	44.63	739	33.93	0.0	0.0
2	2021-12-22	Волинська область	359	115	10.49	3.36	2248	65.71	699	20.43	0.0	0.0
3	2021-12-22	Миколаївська область	251	25	10.07	1.00	1251	50.20	965	38.72	0.0	0.0
4	2021-12-22	Черкаська область	338	122	10.77	3.89	1350	43.03	1327	42.30	0.0	0.0

region_base = alt.Chart(covid_and_vaccination_ukraine_df).mark_area().transform_fold(
    ['Moderna', 'AstraZeneca', 'Pfizer-BioNTech', 'Sinovac (CoronaVac)', 'Johnson & Johnson'],
).encode(
    x=alt.X('month(date):O', title='Month'),
    y=alt.Y('value:Q', aggregate = 'sum', title=None),
    color=alt.Color('key:N', title='Vaccination manufacturer'),
    tooltip=[
        alt.Tooltip('key:N', title='Vaccination manufacturer'),
        alt.Tooltip('value:Q', aggregate='sum', title='Number of people'),
        alt.Tooltip('month(date):O', title='Month'),
   ],
   detail = alt.Detail('region_en:N')
).properties(width=120, height=120)

region_base_title = alt.Chart(covid_and_vaccination_ukraine_df).mark_text(dy = -70).encode(
    text = alt.Text('region_en:N')
)

region_base += region_base_title

region_base.transform_filter(alt.datum.region_en == 'Vinnytsia')

empty_cell = alt.Chart(
    alt.Data(
        values=[{'x': covid_and_vaccination_ukraine_df.date.min(), 'y': 100_000}, {'x': covid_and_vaccination_ukraine_df.date.max(), 'y': 0}]
    )
).mark_point(
    opacity=0
).encode(
    x = alt.X('x:T', axis=alt.Axis(title=None, labelOpacity=0, tickOpacity=0, domainOpacity=0, grid=False)),
    y = alt.Y('y:Q', axis=alt.Axis(title=None, labelOpacity=0, tickOpacity=0, domainOpacity=0, grid=False))
).properties(width=120, height=120)

empty_cell

chart = alt.vconcat()

for row in ukraine_geofaset_grid.row.unique():
    chart_row = alt.hconcat()
    for col in range(1, 9):
        if col in ukraine_geofaset_grid[ukraine_geofaset_grid.row == row].col.unique():
            region_en = ukraine_geofaset_grid[(ukraine_geofaset_grid.row == row) & (ukraine_geofaset_grid.col == col)].name.values[0]
            
            if region_en in covid_stats_ukraine_df.region_en.unique():
                chart_row |= region_base.transform_filter(alt.datum.region_en == region_en)
            else:
                chart_row |= empty_cell
        else:
            chart_row |= empty_cell
    chart &= chart_row

chart.properties(
    title='Vaccination proportions'
).configure_title(
    fontSize=30,
    anchor='middle',
    dy = -20
).configure_view(
    strokeWidth = 0
)

This visualization tries to show the increase of the vaccination trend in the autumn (which is most likely due to new quarantine prohibitions and the increase of new covid case increase shown in the previous sections). Also, this visualization shows a bit unexpected outlier region - Zakarpatska - which is the only region when the 'Johnson & Johnson' vaccination was used.